Numerical estimation of geoelectrical data errors. Comparison with the method of reciprocal measurements.

P. MARTINELLI; F. E. ROBLEDO; P. BORDÓN; A. OSELLA; V. GRUNHUT; M. E. FRANZONI

La Haya

Congreso; 25th European Meeting of Environmental and Engineering Geophysics; 2019

European Association of Geoscientists and Engineers, EAGE

It is well-known that considering appropriate estimations of geoelectrical data errors improves inversion results, especially in the presence of data with atypical values. In these cases, other option is to apply some procedure for removing the outliers from the datasets, and then perform inversions without considering data errors. However, these data are not always easily identifiable.In this work, we present four numerical methods for estimating data errors, alternative to reciprocal measurements. They are based on the analysis of the differences among the data corresponding to neighboring measurement points. They are computationally cheap, and present two important advantages: they do not increase survey times because they do not require any additional measurements, and they can be used with any electrode configuration. Using a representative experimental example, we demonstrate their effectiveness for reducing the effects of outlier data on the inversion results. The subsoil models obtained considering the proposed error estimations are, in general, similar to that obtained considering the errors estimated from direct and r eciprocal measurements. These models are also in agreement with the results obtained without considering data errors, after removing the outlier data. The formulations presented are 2D, but can easily be generalized to three dimensions.